Models, Cost, and Usage Considerations
The assistant model list is fetched from the OpenAI Models API. The create form defaults to gpt-4o-mini, but the list can include models that are unsuitable for the Responses API flow used by the plugin.
A model is the OpenAI engine that produces the response. Changing it can alter answer quality, instruction following, speed, and token cost even when every other SmartSite setting stays the same. SmartSite displays the model IDs returned by OpenAI, but that list is broader than a guarantee of compatibility with the Responses API request used by this plugin.
Start with the smallest safe step: load the assistant form and wait for the model list. Do not consider the task finished before you change one variable at a time before comparing results; this is where the configuration is tested in the context that truly consumes it.
What this feature does and when to use it
Section titled “What this feature does and when to use it”Choose a model that supports the current Responses API and balance quality, latency, and cost with controlled testing.
Use this feature in the following situations:
- You are choosing the model for a new assistant rather than accepting a default without testing.
- You want to compare quality, latency, and token use with a fixed set of representative questions.
- A previously selected model is unavailable or returns an API compatibility error.
Where to find it
Section titled “Where to find it”Before you begin
Section titled “Before you begin”- SmartSite Assistant is installed and activated.
- You are signed in with an account that can manage WordPress options.
Set it up step by step
Section titled “Set it up step by step”- Load the assistant form and wait for the model list.
- Choose a Responses-compatible text model available to the project.
- Save the assistant and activate it.
- Run a fixed set of representative questions.
- Review local response duration and tokens in Analytics, and organization cost data if the Admin key is configured.
- Change one variable at a time before comparing results.
Fields, controls, and important values
Section titled “Fields, controls, and important values”Model choice changes the balance between capability, response time, and cost, but it cannot compensate for vague instructions or missing knowledge. The other values explain which fixed models are used for administrative generation and how usage data is cached. Use them to choose an affordable model that performs well on your real questions, then improve sources and instructions when answers lack website-specific detail.
| Field, control, or status | What SmartSite Assistant does with it | How to use it and why it matters |
|---|---|---|
| Model | Fetched from OpenAI. gpt-4o-mini is the code default for a newly rendered create form. | “Model” affects newly generated output rather than rewriting content that is already saved. Choose a supported value, test representative questions, and have a qualified reviewer check language, terminology, latency, and cost where those differences matter. |
| Metadata suggestion model | The plugin uses gpt-4.1 internally for Knowledge Base metadata suggestions. | This fixed model drafts summaries, topics, FAQs, facts, and relationships for administrator review. Its output can make sources easier to retrieve, but only after a person removes invented, vague, or misleading suggestions. |
| Translation model | The plugin uses gpt-4.1-mini internally for admin text translation. | Administrative translation uses this fixed model rather than the assistant profile’s runtime choice. It can save rewriting time, although fluent review is still needed because terminology errors may change instructions or visitor expectations. |
| Usage cache | OpenAI usage/cost results are cached for one hour; a stale cached result may be shown if refresh fails. | Choose “Usage cache” from observed needs rather than guesswork. Boundary values can affect availability, detection, display, or reporting; test a normal case and an edge case so useful conversations are not sacrificed for an arbitrary number. |
How to confirm it is working
Section titled “How to confirm it is working”Use a separate test session to confirm Models, Cost, and Usage Considerations. This keeps existing login, browser storage, and response history from hiding the change, and it shows whether the result reaches the complete workflow rather than stopping at WordPress storage.
Practical example
Section titled “Practical example”Compare five support questions using the default model before choosing a larger model solely for perceived quality.
Recommended practice
Section titled “Recommended practice”- Change one part of Models, Cost, and Usage Considerations at a time and keep a short record of the previous value and test result.
- Verify the saved result in the screen, visitor session, or connected service that actually consumes the setting.
Important warnings
Section titled “Important warnings”Common problems and focused checks
Section titled “Common problems and focused checks”| Problem | What to check and what to do next |
|---|---|
| A selected model appears in the list but chat fails. | Edit the assistant and select a model known to support the Responses API for your project. |
| Models, Cost, and Usage Considerations is missing or does not match this guide. | Confirm the plugin is active and the account can manage WordPress options. Separate credential, project, model, and network failures before changing assistant content. |
| A change on Models, Cost, and Usage Considerations does not produce the expected result. | Keep the exact notice and test case, then review the browser console and WordPress/PHP log. Separate credential, project, model, and network failures before changing assistant content. |
Screen reference
Section titled “Screen reference”- Capture
- Show the assistant Model selector open with the current selection visible; avoid implying that every listed model is supported.
- Show
- Model label, selected model, save action
- Viewport
- Desktop, 1440 × 900
- Annotate
- Use numbered callouts only for controls referenced in the procedure.
- Redact
- OpenAI keys, tokens, secrets, personal information, private URLs, IP addresses, and conversation text